Changed to pageviews instead of unique pageviews
--- a/ckanext/ga_report/controller.py
+++ b/ckanext/ga_report/controller.py
@@ -22,6 +22,7 @@
def _month_details(cls):
+ '''Returns a list of all the month names'''
months = []
vals = model.Session.query(cls.period_name).filter(cls.period_name!='All').distinct().all()
for m in vals:
@@ -243,11 +244,11 @@
if publisher:
q = q.filter(GA_Url.department_id==publisher.name)
q = q.filter(GA_Url.period_name==month)
- q = q.order_by('ga_url.visitors::int desc')
+ q = q.order_by('ga_url.visits::int desc')
top_packages = []
for entry,package in q.limit(count):
if package:
- top_packages.append((package, entry.pageviews, entry.visitors))
+ top_packages.append((package, entry.pageviews, entry.visits))
else:
log.warning('Could not find package associated package')
@@ -305,9 +306,11 @@
month = c.month or 'All'
connection = model.Session.connection()
q = """
- select department_id, sum(pageviews::int) views, sum(visitors::int) visits
+ select department_id, sum(pageviews::int) views, sum(visits::int) visits
from ga_url
where department_id <> ''
+ and package_id <> ''
+ and url like '/dataset/%%'
and period_name=%s
group by department_id order by visits desc
"""
--- a/ckanext/ga_report/download_analytics.py
+++ b/ckanext/ga_report/download_analytics.py
@@ -92,33 +92,40 @@
def download_and_store(self, periods):
for period_name, period_complete_day, start_date, end_date in periods:
+ log.info('Period "%s" (%s - %s)',
+ self.get_full_period_name(period_name, period_complete_day),
+ start_date.strftime('%Y-%m-%d'),
+ end_date.strftime('%Y-%m-%d'))
+
if self.delete_first:
- log.info('Deleting existing Analytics for period "%s"',
+ log.info('Deleting existing Analytics for this period "%s"',
period_name)
ga_model.delete(period_name)
- log.info('Downloading Analytics for period "%s" (%s - %s)',
- self.get_full_period_name(period_name, period_complete_day),
- start_date.strftime('%Y %m %d'),
- end_date.strftime('%Y %m %d'))
# Clean up the entries before we run this
ga_model.pre_update_url_stats(period_name)
accountName = config.get('googleanalytics.account')
+ log.info('Downloading analytics for dataset views')
data = self.download(start_date, end_date, '~/%s/dataset/[a-z0-9-_]+' % accountName)
- log.info('Storing Dataset Analytics for period "%s"',
- self.get_full_period_name(period_name, period_complete_day))
+
+ log.info('Storing dataset views (%i rows)', len(data.get('url')))
self.store(period_name, period_complete_day, data, )
+ log.info('Downloading analytics for publisher views')
data = self.download(start_date, end_date, '~/%s/publisher/[a-z0-9-_]+' % accountName)
- log.info('Storing Publisher Analytics for period "%s"',
- self.get_full_period_name(period_name, period_complete_day))
+
+ log.info('Storing publisher views (%i rows)', len(data.get('url')))
self.store(period_name, period_complete_day, data,)
+ log.info('Aggregating datasets by publisher')
ga_model.update_publisher_stats(period_name) # about 30 seconds.
+
+ log.info('Downloading and storing analytics for site-wide stats')
self.sitewide_stats( period_name )
+ log.info('Downloading and storing analytics for social networks')
self.update_social_info(period_name, start_date, end_date)
@@ -152,8 +159,8 @@
start_date = start_date.strftime('%Y-%m-%d')
end_date = end_date.strftime('%Y-%m-%d')
query = 'ga:pagePath=%s$' % path
- metrics = 'ga:uniquePageviews, ga:visits'
- sort = '-ga:uniquePageviews'
+ metrics = 'ga:pageviews, ga:visits'
+ sort = '-ga:pageviews'
# Supported query params at
# https://developers.google.com/analytics/devguides/reporting/core/v3/reference
@@ -187,12 +194,10 @@
start_date = '%s-01' % period_name
end_date = '%s-%s' % (period_name, last_day_of_month)
- print 'Sitewide_stats for %s (%s -> %s)' % (period_name, start_date, end_date)
-
funcs = ['_totals_stats', '_social_stats', '_os_stats',
'_locale_stats', '_browser_stats', '_mobile_stats']
for f in funcs:
- print ' + Fetching %s stats' % f.split('_')[1]
+ log.info('Downloading analytics for %s' % f.split('_')[1])
getattr(self, f)(start_date, end_date, period_name)
def _get_results(result_data, f):
@@ -207,8 +212,8 @@
results = self.service.data().ga().get(
ids='ga:' + self.profile_id,
start_date=start_date,
- metrics='ga:uniquePageviews',
- sort='-ga:uniquePageviews',
+ metrics='ga:pageviews',
+ sort='-ga:pageviews',
max_results=10000,
end_date=end_date).execute()
result_data = results.get('rows')
@@ -237,7 +242,7 @@
ids='ga:' + self.profile_id,
filters='ga:pagePath=~%s$' % (path,),
start_date=start_date,
- metrics='ga:bounces,ga:uniquePageviews',
+ metrics='ga:bounces,ga:pageviews',
dimensions='ga:pagePath',
max_results=10000,
end_date=end_date).execute()
@@ -255,8 +260,8 @@
results = self.service.data().ga().get(
ids='ga:' + self.profile_id,
start_date=start_date,
- metrics='ga:uniquePageviews',
- sort='-ga:uniquePageviews',
+ metrics='ga:pageviews',
+ sort='-ga:pageviews',
dimensions="ga:language,ga:country",
max_results=10000,
end_date=end_date).execute()
@@ -279,8 +284,8 @@
results = self.service.data().ga().get(
ids='ga:' + self.profile_id,
start_date=start_date,
- metrics='ga:uniquePageviews',
- sort='-ga:uniquePageviews',
+ metrics='ga:pageviews',
+ sort='-ga:pageviews',
dimensions="ga:socialNetwork,ga:referralPath",
max_results=10000,
end_date=end_date).execute()
@@ -298,8 +303,8 @@
results = self.service.data().ga().get(
ids='ga:' + self.profile_id,
start_date=start_date,
- metrics='ga:uniquePageviews',
- sort='-ga:uniquePageviews',
+ metrics='ga:pageviews',
+ sort='-ga:pageviews',
dimensions="ga:operatingSystem,ga:operatingSystemVersion",
max_results=10000,
end_date=end_date).execute()
@@ -323,8 +328,8 @@
results = self.service.data().ga().get(
ids='ga:' + self.profile_id,
start_date=start_date,
- metrics='ga:uniquePageviews',
- sort='-ga:uniquePageviews',
+ metrics='ga:pageviews',
+ sort='-ga:pageviews',
dimensions="ga:browser,ga:browserVersion",
max_results=10000,
end_date=end_date).execute()
@@ -372,8 +377,8 @@
results = self.service.data().ga().get(
ids='ga:' + self.profile_id,
start_date=start_date,
- metrics='ga:uniquePageviews',
- sort='-ga:uniquePageviews',
+ metrics='ga:pageviews',
+ sort='-ga:pageviews',
dimensions="ga:mobileDeviceBranding, ga:mobileDeviceInfo",
max_results=10000,
end_date=end_date).execute()
--- a/ckanext/ga_report/ga_model.py
+++ b/ckanext/ga_report/ga_model.py
@@ -27,7 +27,7 @@
Column('period_name', types.UnicodeText),
Column('period_complete_day', types.Integer),
Column('pageviews', types.UnicodeText),
- Column('visitors', types.UnicodeText),
+ Column('visits', types.UnicodeText),
Column('url', types.UnicodeText),
Column('department_id', types.UnicodeText),
Column('package_id', types.UnicodeText),
@@ -63,7 +63,7 @@
Column('period_name', types.UnicodeText),
Column('publisher_name', types.UnicodeText),
Column('views', types.UnicodeText),
- Column('visitors', types.UnicodeText),
+ Column('visits', types.UnicodeText),
Column('toplevel', types.Boolean, default=False),
Column('subpublishercount', types.Integer, default=0),
Column('parent', types.UnicodeText),
@@ -116,7 +116,7 @@
return '/' + '/'.join(url.split('/')[2:])
-def _get_department_id_of_url(url):
+def _get_package_and_publisher(url):
# e.g. /dataset/fuel_prices
# e.g. /dataset/fuel_prices/resource/e63380d4
dataset_match = re.match('/dataset/([^/]+)(/.*)?', url)
@@ -126,12 +126,13 @@
if dataset:
publisher_groups = dataset.get_groups('publisher')
if publisher_groups:
- return publisher_groups[0].name
+ return dataset_ref,publisher_groups[0].name
+ return dataset_ref, None
else:
publisher_match = re.match('/publisher/([^/]+)(/.*)?', url)
if publisher_match:
- return publisher_match.groups()[0]
-
+ return None, publisher_match.groups()[0]
+ return None, None
def update_sitewide_stats(period_name, stat_name, data):
for k,v in data.iteritems():
@@ -156,25 +157,6 @@
model.Session.commit()
-def update_url_stat_totals(period_name):
-
- """
- items = model.Session.query(GA_Url).\
- filter(GA_Url.period_name != "All").\
- filter(GA_Url.url==url).all()
- values = {'id': make_uuid(),
- 'period_name': "All",
- 'period_complete_day': "0",
- 'url': url,
- 'pageviews': sum([int(x.pageviews) for x in items]),
- 'visitors': sum([int(x.visitors) for x in items]),
- 'department_id': department_id,
- 'package_id': package
- }
- model.Session.add(GA_Url(**values))
- model.Session.commit()
- """
-
def pre_update_url_stats(period_name):
model.Session.query(GA_Url).\
filter(GA_Url.period_name==period_name).delete()
@@ -184,26 +166,42 @@
def update_url_stats(period_name, period_complete_day, url_data):
- for url, views, visitors in url_data:
- department_id = _get_department_id_of_url(url)
-
- package = None
- if url.startswith('/dataset/'):
- package = url[len('/dataset/'):]
-
- values = {'id': make_uuid(),
- 'period_name': period_name,
- 'period_complete_day': period_complete_day,
- 'url': url,
- 'pageviews': views,
- 'visitors': visitors,
- 'department_id': department_id,
- 'package_id': package
- }
- model.Session.add(GA_Url(**values))
+ for url, views, visits in url_data:
+ package, publisher = _get_package_and_publisher(url)
+
+
+ item = model.Session.query(GA_Url).\
+ filter(GA_Url.period_name==period_name).\
+ filter(GA_Url.url==url).first()
+ if item:
+ item.pageviews = item.pageviews + views
+ item.visits = item.visits + visits
+ if not item.package_id:
+ item.package_id = package
+ if not item.department_id:
+ item.department_id = publisher
+ model.Session.add(item)
+ else:
+ values = {'id': make_uuid(),
+ 'period_name': period_name,
+ 'period_complete_day': period_complete_day,
+ 'url': url,
+ 'pageviews': views,
+ 'visits': visits,
+ 'department_id': publisher,
+ 'package_id': package
+ }
+ model.Session.add(GA_Url(**values))
model.Session.commit()
if package:
+ old_pageviews, old_visits = 0, 0
+ old = model.Session.query(GA_Url).\
+ filter(GA_Url.period_name=='All').\
+ filter(GA_Url.url==url).all()
+ old_pageviews = sum([int(o.pageviews) for o in old])
+ old_visits = sum([int(o.visits) for o in old])
+
entries = model.Session.query(GA_Url).\
filter(GA_Url.period_name!='All').\
filter(GA_Url.url==url).all()
@@ -211,14 +209,14 @@
'period_name': 'All',
'period_complete_day': 0,
'url': url,
- 'pageviews': sum([int(e.pageviews) for e in entries]),
- 'visitors': sum([int(e.visitors) for e in entries]),
- 'department_id': department_id,
+ 'pageviews': sum([int(e.pageviews) for e in entries]) + old_pageviews,
+ 'visits': sum([int(e.visits) for e in entries]) + old_visits,
+ 'department_id': publisher,
'package_id': package
}
+
model.Session.add(GA_Url(**values))
model.Session.commit()
-
@@ -262,7 +260,7 @@
filter(model.Group.type=='publisher').\
filter(model.Group.state=='active').all()
for publisher in publishers:
- views, visitors, subpub = update_publisher(period_name, publisher, publisher.name)
+ views, visits, subpub = update_publisher(period_name, publisher, publisher.name)
parent, parents = '', publisher.get_groups('publisher')
if parents:
parent = parents[0].name
@@ -271,7 +269,7 @@
filter(GA_Publisher.publisher_name==publisher.name).first()
if item:
item.views = views
- item.visitors = visitors
+ item.visits = visits
item.publisher_name = publisher.name
item.toplevel = publisher in toplevel
item.subpublishercount = subpub
@@ -283,7 +281,7 @@
'period_name': period_name,
'publisher_name': publisher.name,
'views': views,
- 'visitors': visitors,
+ 'visits': visits,
'toplevel': publisher in toplevel,
'subpublishercount': subpub,
'parent': parent
@@ -293,7 +291,7 @@
def update_publisher(period_name, pub, part=''):
- views,visitors,subpub = 0, 0, 0
+ views,visits,subpub = 0, 0, 0
for publisher in go_down_tree(pub):
subpub = subpub + 1
items = model.Session.query(GA_Url).\
@@ -301,9 +299,9 @@
filter(GA_Url.department_id==publisher.name).all()
for item in items:
views = views + int(item.pageviews)
- visitors = visitors + int(item.visitors)
-
- return views, visitors, (subpub-1)
+ visits = visits + int(item.visits)
+
+ return views, visits, (subpub-1)
def get_top_level():
--- a/ckanext/ga_report/helpers.py
+++ b/ckanext/ga_report/helpers.py
@@ -1,7 +1,9 @@
import logging
import operator
+
import ckan.lib.base as base
import ckan.model as model
+from ckan.logic import get_action
from ckanext.ga_report.ga_model import GA_Url, GA_Publisher
from ckanext.ga_report.controller import _get_publishers
@@ -39,25 +41,38 @@
order_by('ga_url.pageviews::int desc')
num_top_datasets = top_datasets.count()
+ dataset = None
if num_top_datasets:
- dataset = None
+ count = 0
while not dataset:
rand = random.randrange(0, min(top, num_top_datasets))
ga_url = top_datasets[rand]
dataset = model.Package.get(ga_url.url[len('/dataset/'):])
if dataset and not dataset.state == 'active':
dataset = None
- else:
+ count += 1
+ if count > 10:
+ break
+ if not dataset:
+ # fallback
dataset = model.Session.query(model.Package)\
.filter_by(state='active').first()
- publisher = dataset.get_groups('publisher')[0]
- return {
- 'dataset': dataset,
- 'publisher': publisher
- }
+ if not dataset:
+ return None
+ dataset_dict = get_action('package_show')({'model': model,
+ 'session': model.Session},
+ {'id':dataset.id})
+ return dataset_dict
def single_popular_dataset_html(top=20):
- context = single_popular_dataset(top)
+ dataset_dict = single_popular_dataset(top)
+ groups = package.get('groups', [])
+ publishers = [ g for g in groups if g.get('type') == 'publisher' ]
+ publisher = publishers[0] if publishers else {'name':'', 'title': ''}
+ context = {
+ 'dataset': dataset_dict,
+ 'publisher': publisher_dict
+ }
return base.render_snippet('ga_report/ga_popular_single.html', **context)
@@ -90,7 +105,7 @@
if not p in datasets:
datasets[p] = {'views':0, 'visits': 0}
datasets[p]['views'] = datasets[p]['views'] + int(entry.pageviews)
- datasets[p]['visits'] = datasets[p]['visits'] + int(entry.visitors)
+ datasets[p]['visits'] = datasets[p]['visits'] + int(entry.visits)
results = []
for k, v in datasets.iteritems():
--- a/ckanext/ga_report/templates/ga_report/ga_popular_single.html
+++ b/ckanext/ga_report/templates/ga_report/ga_popular_single.html
@@ -8,14 +8,14 @@
<h2>Featured dataset</h2>
<div class="dataset-summary boxed">
- <a class="dataset-header" href="${h.url_for(controller='package', action='read', id=dataset.name)}">
- <h3>${dataset.title}</h3>
+ <a class="dataset-header" href="${h.url_for(controller='package', action='read', id=dataset['name'])}">
+ <h3>${dataset['title']}</h3>
</a>
<h4>
<strong>Publisher</strong> :
- <a href="/publisher/${publisher.name}">${publisher.title}</a>
+ <a href="/publisher/${publisher['name']}">${publisher['title']}</a>
</h4>
- <div>${h.truncate(dataset.notes, length=200, whole_word=True)}</div>
+ <div>${h.truncate(dataset['notes_rendered'], length=200, whole_word=True)}</div>
</div>
<div>
<a href="${h.url_for(controller='ckanext.ga_report.controller:GaDatasetReport',action='read')}" class="btn">Other popular datasets</a>
--- a/ckanext/ga_report/templates/ga_report/notes.html
+++ b/ckanext/ga_report/templates/ga_report/notes.html
@@ -6,11 +6,10 @@
<li class="widget-container boxed widget_text">
<h4>Notes</h4>
<ul>
- <li>'Views' is the number of sessions during which that page was viewed one or more times ('Unique Pageviews').</li>
-<!-- <li>'Visits' is the number of individual sessions initiated by all the visitors to your site, counted once for each visitor for each session.</li>-->
- <li>'Visitors' is the number of unique users visiting the site (whether once or more times).</li>
+ <li>"Views" is the number of sessions during which the page was viewed one or more times (technically known as "unique pageviews").</li>
+ <li>"Visits" is the number of unique user visits to a page, counted once for each visitor for each session.</li>
<li>These usage statistics are confined to users with javascript enabled, which excludes web crawlers and API calls.</li>
- <li>The results for only small numbers of views/visits are not shown. Where these relate to site pages, then they are available in full in the CSV download. Where these relate to users' web browser information, they are not disclosed, for privacy reasons.</li>
+ <li>The results are not shown when the number of views/visits is tiny. Where these relate to site pages, results are available in full in the CSV download. Where these relate to users' web browser information, results are not disclosed, for privacy reasons.</li>
</ul>
</li>
</html>